How data, AI, and intelligent technologies are transforming electricity grids

For decades, this worked well. Energy demand was predictable, and utilities burned enough coal and gas to meet it – a bit less on long summer days, a bit more on short winter ones, or when the national team reaches the cup final.

But things have changed, and this way of doing things is no longer sufficient. Reliable coal and gas furnaces are out, and intermittent solar and wind are in. Demand is changing rapidly and unevenly, thanks to the electrification of transport, heating, and industrial processes, as well as to extreme weather that makes hot days hotter and cold days colder.

All of this transforms the grid from means of carrying energy from power stations to consumers, into a complex, dynamic, marketplace for energy. Aging infrastructure – that was not designed for decentralized energy – doesn’t help matters.

Reinventing the grid for a decentralized, decarbonized world

Utilities face several challenges all at once.

They need to grow capacity to meet surging electricity demand – our own estimates suggest 125 million kilometers of transmission and distribution lines are needed in the next 30 years (up from 80 million km today), at a cost of $7trn per year^[1]. To deliver that efficiently, they need to become much better at predicting short- and long-term demand, so they can make intelligent decisions.

They need to rethink the grid around decentralized and intermittent renewable energy inputs, from huge wind farms to dispersed rooftop solar. In doing so, they will need to deal with a complicated mix of new and aging assets for generation, transmission & distribution, and storage.

Increasingly, they also must offer consumer services, as end users now expect more detailed data on energy usage and billing, and tools to sell surplus energy back to the grid.

With the increasing digitization and interconnectedness of the electric grid, cybersecurity has become a paramount concern. The grid’s vulnerability to cyber threats poses significant challenges, as malicious actors target critical infrastructure, aiming to disrupt operations, cause widespread outages, and compromise the reliability and safety of the energy system.

To compound matters, utilities are faced with the challenge of navigating a complex and evolving regulatory environment. New policies, mandates, and standards are being introduced to address emerging issues such as environmental sustainability, grid modernization, and cybersecurity.

All of this is leading utilities to ask new questions: How can we make the most of decentralized energy sources? How can we optimize aging assets? How can we accurately predict supply and demand in this more complex world? How can we improve the customer experience? And how can we keep the energy system secure?

To navigate these challenges successfully, utilities must define a vision and design a clear, tailored roadmap step by step, with clear added value and risk reduction regarding the constraints they will face. This strategic approach is critical, not only for adapting to the evolving energy landscape, but also for ensuring the resilience and efficiency of the grid in the face of rapid technological and environmental changes.

To create the grid of the future – and so answer all these questions – we need to do more with data and AI.

Making intelligent decisions

The heart of this transformation is about using data to generate situational awareness of energy infrastructure, so utilities can make intelligent decisions.

Take EV ownership. The transition to EVs will place massive new electricity demands on the grid. But when, and where, and how fast? Will everyone charge slowly overnight, or quickly at lunchtime? Can people be incentivized to charge at quiet times? We need to know these things to decide how much more distribution capacity to build or upgrade, and when.

That needs clever models to predict the future, and such models need a wide range of data. Utilities may already have some of that, such as electricity usage amongst existing EV drivers. But other data – such as EV sales projections, the percentage of people with driveways, public charging infrastructure plans – will need to be sourced from elsewhere. That is something new for utilities.

Perhaps the most critical models will be those for balancing renewable energy. By combining accurate weather models (e.g. how much sun will shine, and wind will blow) with a digital twin of your generation infrastructure, you can predict how much renewable energy will be fed in on a given day. By building precise energy demand models – e.g., using smart meter data, historical usage data, and weather data – you can predict how much energy is needed on that same day.

The gap between the minimum predicted renewable production, and maximum predicted energy demand, tells you how much fossil fuels you need to burn to ensure the lights stay on. The more accurate the models, the less you need to burn.

These are two of many examples of how data will transform the grid. Others include models that nudge consumers to use electricity at different times, models to predict sudden energy demand spikes, predictive models for asset health and vegetation management (asset failures can cost $millions in downtime, so are best avoided).

Technology drives grid transformation

All this will require wholesale transformation that builds intelligence into energy systems, turning utilities into companies that routinely use high-quality data to develop and deploy models – from load balancing to infrastructure planning, to predictive asset maintenance.

That will need people, processes, and technology infrastructure.

It will need people who can combine high-quality data from multiple sources to build highly predictive models and generate actionable insights for the company and its customers. This is not just about modeling, but also making smart decisions about data, such as where to focus limited resources, what data sources to acquire and use, and whether to build AI tools in the cloud or at the edge.

It will need processes to gather data. That will mean changes to your own data sources – e.g., by deploying smart meters to gather data, and adding connected assets (smart new ones or retrofitting aging ones with practical sensors) to monitor performance and build a cohesive model of the grid. It will need new relationships to secure data from third party sources – from weather companies to EV sales analysts, to government electric heating installation programs.

And utilities companies will need to build the IT infrastructure backbone that securely collects data from these many sources and transports it into a shared cloud platform. It will need tools to aggregate disparate data into consistent formats that can be used to build new models and feed existing ones. And it will need to deliver those insights – via purpose-built digital interfaces – to the people who need to act on them, whether network planners, asset maintenance engineers, or energy users.

Conclusion

Better technology for data collection and model-building (both AI and classical), will be critical to transforming the grid into one that is fit for the future. Technology is often thought of as an enabler of change, but in this case, that is thinking too small. Technology is the driver of change. It is the only way to create a smart grid that will deliver the decentralized, decarbonized energy system we need.

---

^[1] World Energy Markets Observatory 2023, Capgemini,